To investigate associations between stress, resilience, and burnout in three emotional job sectors.
We conducted a multi-group comparative study using structural equation modeling and latent mean analysis. In total, 806 participants (403 call center consultants, 270 mental health workers, and 133 school counselors) completed self-administered questionnaires including the Perceived Stress Scale, Korean version of the Connor-Davidson Resilience Scale, and Maslach Burnout Inventory-General Survey.
Stress had significant direct effects on resilience and burnout, and resilience had significant direct effects on burnout in all groups. Resilience partially mediated these relationships among call center consultants and school counselors. Stress and burnout were highest in call center consultants, followed, in order, by mental health workers and school counselors. Resilience was highest in school counselors, followed, in order, by mental health workers and call center consultants. The effect size of the latent mean difference was highest for burnout, followed, in order, by resilience and stress.
Our findings suggest that stress caused by emotional labor can contribute to burnout. Interventions targeted at different sectors are needed to reduce burnout.
As the scope of human service fields has expanded, the expression of emotions, such as hospitality and kindness, have emerged as critical in service industries [
Among call center consultants, psychological problems rooted in the nature of their job are increasing; a call center consultant is frequently exposed to anger, hostility, and verbal violence, in addition to the basic expectations of consultants, which are to provide customers with information and meet their needs [
Therefore, it is critical to understand the seriousness of the problems resulting from emotional labor and to provide interventions to prevent burnout. Differentiated intervention strategies should also be identified. However, comparative studies of different jobs involving emotional labor have been absent. Thus, this research intends to extend the scope of previous research examining simple linear correlations with respect to each emotional job sector, by conducting a multigroup comparative study specifically on call center consultants, mental health workers, and school counselors. Specifically, burnout, stress, and resilience were selected as major variables, because stress is reported as a risk factor for burnout [
Cross-sectional surveys were conducted from September 2015 to May 2016 by researchers from the Gwangju Mental Health Commission and Gwangju Metropolitan Mental Health and Welfare Center, who visited participants’ work or training venues. Self-administered questionnaires were delivered to 403 call center consultants working at 1 of 12 customer service centers, 270 mental health professionals working at community mental health centers, and 133 school counselors in Gwangju. All participants completed the surveys, which were filled out anonymously to protect respondents’ privacy. The study was approved by the Institutional Review Board of Chonnam National University (CNUH-2015-171; 172).
Perceived stress was measured using the Perceived Stress Scale, which consists of ten items measuring specific feelings experienced in the past one month [
Resilience was measured using the Korean version of the Connor-Davidson Resilience Scale (K-CD-RISC), which consists of 25 items measuring resilience as an ability to cope with stress successfully [
Burnout was measured using the Maslach Burnout Inventory-General Survey (MBI-GS), which contains three dimensions: exhaustion, cynicism, and professional efficacy [
To analyze the data for this research, SPSS 20.0 and AMOS 20.0 (SPSS Inc., Chicago, IL, USA) were used. Exploratory and confirmatory factor analyses were conducted. To measure the scales’ reliability, Cronbach’s α values were calculated. Descriptive statistics, such as frequencies, means, standard deviations, and normality analyses, were also derived. For the comparative analysis of direct and indirect effects, SEM was conducted. To assess the goodness of fit, indices such as
The socio-demographic characteristics of participants are described in
About 72% of call center consultants received verbal threats from clients and 30% experienced sexual harassment from clients. Although 65.8% of consultants felt the need to receive counseling from a psychological counselor, professional counselors were available only in 14.1% of centers. Among mental health workers, 72.6% experienced verbal threats, 33.7% received physical threats, and 18.9% experienced sexual harassment from patients. Furthermore, 33.3% suffered from emotional sequelae from patients’ suicide attempts. Among school counselors, 33.8% experienced verbal threats, 15.8% had physical threats, and 12.0% experienced sexual harassment by students. Furthermore, 27.8% experienced verbal threats from guardians, and 53.4% suffered from emotional sequelae from students’ suicide attempts.
In SEM, if measurement variables do not have normal distributions, the assumption of multivariate normality is not met. The wrong estimates produced prevent proper statistical testing. Considering the conditions of normality (skewness lower than 2, kurtosis lower than 7) in SEM [
In the research, we tested goodness of fit based on CFI, TLI, and RMSEA, which are not sensitive to sample size and have established evaluation criteria that consider simplicity. The goodness of fit of the study model was satisfactory for all indices, except for
In all groups (we had three groups in the survey: call center consultants, mental health workers, and school counselors), the results indicated that stress had a significant direct effect on resilience and burnout, and resilience had a significant direct effect on burnout (
Gender, marital status, age, and educational level were controlled. These control variables had some significant influence on the major variables (
Significance testing of indirect effects was performed using the bootstrapping method. The analysis of the stress-resilience-burnout path showed that for the call center consultants (β=0.138, p=0.002) and school counselors models (β=0.147, p=0.008), the partial mediating effect of resilience was significant, but for the mental health workers model (β=0.084, p=0.116), the indirect effect of resilience was not significant (
LMA has an advantage of taking measurement errors into account between each variable compared with ANOVA, which directly compares the variables [
The first validation is for configural invariance. This is a step to validate if measurement variables load on the same factors for each group. In both comparative models, configural invariance was validated at a significant level (
The second step is the validation of metrics invariance. We compared the
The third step is the validation of scalar invariance. In scalar invariance, invariance restrictions are placed on the intercepts of each measurement variable. As given in
As the mean value of factors cannot be directly compared in LMA, the latent mean of a reference group should be fixed to 0 to predict the latent means of other groups [
Moreover, Cohen’s effect sizes (d) [
In the mental health related worker-call center consultant comparative model, while the call center consultants’ stress and burnout were higher than those of the mental health workers, their resilience turned out to be low. The effect sizes indicated that stress was moderate, resilience was high, and burnout was considerably high. In the mental health worker-school counselor comparative model, while the mental health workers’ stress and burnout levels were generally higher than those of the school counselors, the mental health workers’ resilience was lower. The effect sizes showed that stress was moderate, whereas resilience and burnout were high. In the school counselor-call center consultant comparative model, compared with school counselors, stress and burnout were relatively high for call center consultants, whereas their resilience was low. The effect sizes described the stress as high, and resilience and burnout as considerably high (
This research conducted SEM analysis and LMA on the stress, resilience, and burnout in three job sectors known to involve high levels of emotional labor. Our results confirmed that stress had significant direct effects on resilience and burnout, and resilience had significant direct effects on burnout in all groups. The findings support previous research claiming that stress caused by emotional labor can be a cause of burnout [
Our results are related to the indirect effects of resilience in the relationship between stress and burnout. Partial mediating effects of resilience were found in the call center consultant and school counselor groups, and the indirect effect was not significant in the mental health worker group. The call center consultant case supports previous research results showing that resilience has a mediating effect [
For mental health workers, the indirect effect of resilience in the relationship between stress and burnout was shown to be insignificant. This outcome is a divergent finding compared with previous studies on clinical nurses, social workers, and public workers with respect to the indirect effects of resilience found in the relationship between emotional labor stress and burnout [
Moreover, in burnout research on psychiatric ward social workers, who have similar emotional labor experiences to the mental health workers in this research, the multiple mediating effects of resilience and social support were proven to be significant [
Our results are related to the LMA of the major variables. In terms of stress and burnout levels, stress and burnout were highest in call center consultants, followed, in order, by mental health workers and school counselors. Resilience was highest in school counselors, followed, in order, by mental health workers and call center consultants. The effect size of the latent mean difference was highest for burnout, followed, in order, by resilience and stress. Emotional labor showed discrepancies in stress, resilience, and burnout levels across different job sectors.
Call center consultants had the most stress and burnout, and the lowest resilience. This finding is consistent with the results of existing research reporting that this job involves the most emotional labor [
Mental health workers had relatively lower stress and burnout levels and higher resilience compared with call center consultants, but had higher stress and burnout and lower resilience than school counselors. Mental health workers are exposed to active supervision systems and supplementary training. The results indicate that support to reduce job-related stress and burnout and to promote protective factors remains unsatisfactory [
As a high proportion of workers in these sectors reported experiencing verbal and physical threats at work, their mental health status should be assessed periodically, and professional counseling services should be provided at their workplace. Moreover, governmental efforts to promote job mental health services are crucial. The enactment of the Emotional Labor Worker Protection Act is urgently needed, and call center consultants should be included as a high-risk group for mental health problems, so that systematic adjustments can be made accordingly. In addition, it is also important to develop a job mental health manual for call center workers and provide mental health services in collaboration with regional and basic centers. Trauma prevention and intervention programs should be implemented for mental health workers and school counselors due to clients’ violence or suicidal attempts. Psychological support such as debriefing trauma, various coping strategies, and resilience promotion programs should be activated.
The research has the following limitations. First, as the study was limited to a certain area, there are external validity limitations to generalizing the results to a national level. In future research, the study area should be expanded. Also this study is limited in that the gender ratio was not balanced with about 90% of the study participants being female. It is imperative to remedy this by collecting the samples of male emotional laborers so that gender-based analysis of the psychological characteristics of emotional laborers can be performed in depth. Second, the burnout levels of the subjects in this study were higher than the results of previous studies that used the same scale [
The online-only Data Supplement is available with this article at
This study was conducted as part of the Gwangju Mental Health Demonstration Project supported by Ministry of Health & Welfare and Gwangju Metropolitan city, Republic of Korea. This study was supported by Nambu University.
The authors have no potential conflicts of interest to disclose.
Conceptualization: Ju-Yeon Lee, Sung-Wan Kim. Data curation: Il-Seon Shin, Jin-Sang Yoon, Sung-Wan Kim. Formal analysis: Yu-Ri Lee. Funding acquisition: Sung-Wan Kim. Methodology: Ju-Yeon Lee, Yu-Ri Lee. Project administration: Yu-Ri Lee, Ju-Yeon Lee, Sung-Wan Kim. Supervision: Jae-Min Kim, Il-Seon Shin, Jin-Sang Yoon. Writing—original draft: Yu-Ri Lee. Writing—review & editing: Ju-Yeon Lee, Jae-Min Kim, Il-Seon Shin, Jin-Sang Yoon, Sung-Wan Kim.
Study model. e: measurement error, d: unexplained error.
Sociodemographic characteristics of emotional laborers
Call center workers (N=403) | Mental health workers (N=270) | School counselors (N=133) | |
---|---|---|---|
Gender (%) | |||
Male | 28 (7.0) | 56 (20.7) | 7 (5.3) |
Female | 375 (93.0) | 214 (79.3) | 126 (94.7) |
Age, years (%) | |||
20–30 | 67 (16.6) | 82 (30.3) | 17 (12.8) |
31–40 | 242 (60.1) | 112 (41.6) | 22 (15.5) |
≥41 | 94 (23.3) | 76 (28.1) | 94 (70.7) |
Marital status (%) | |||
Married | 261 (64.8) | 151 (51.9) | 108 (81.2) |
Not married | 142 (35.2) | 119 (44.1) | 25 (18.8) |
Educational level (%) | |||
High school | 92 (22.9) | - | - |
University | 308 (76.4) | 211 (78.1) | 46 (34.6) |
Graduate school | 3 (0.7) | 59 (21.9) | 87 (65.4) |
Estimates of study model and mediating effects
Pathway | Call-center workers |
Mental health workers |
School counselors |
||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
B | β | SE | CR | B | β | SE | CR | B | β | SE | CR | ||
Stress → resilience | -0.340 | -0.401 | 0.049 | -6.891 |
-0.521 | -0.497 | 0.077 | -6.775 |
-0.457 | -0.439 | 0.113 | -4.037 |
|
Resilience → burnout | -2.669 | -0.343 | 0.464 | -5.756 |
-1.226 | -0.168 | 0.556 | -2.206 |
-3.202 | -0.335 | 1.034 | -3.097 |
|
Stress → burnout | 3.103 | 0.470 | 0.402 | 7.727 |
5.297 | 0.693 | 0.703 | 7.530 |
5.577 | 0.560 | 1.145 | 4.868 |
|
Gender → stress | -0.297 | -0.109 | 0.146 | -2.030 |
-0.132 | -0.085 | 0.105 | -1.259 | -0.150 | -0.060 | 0.231 | -0.651 | |
Age → stress | 0.080 | 0.043 | 0.113 | 0.705 | 0.208 | 0.085 | 0.113 | 1.040 | 0.091 | 0.074 | 0.149 | 0.607 | |
Marital status → stress | 0.013 | 0.009 | 0.087 | 0.153 | 0.117 | 0.163 | 0.101 | 2.056 |
0.423 | 0.295 | 0.180 | 2.351 |
|
Educational level → stress | 0.120 | 0.073 | 0.090 | 1.334 | 0.099 | 0.065 | 0.107 | 0.927 | -0.106 | -0.090 | 0.108 | -0.984 | |
Gender → resilience | 0.180 | 0.078 | 0.118 | 1.527 | 0.005 | 0.003 | 0.095 | 0.055 | -0.233 | -0.089 | 0.219 | -1.066 | |
Age → resilience | -0.145 | -0.092 | 0.091 | -1.589 | -0.008 | -0.005 | 0.102 | -0.075 | -0.425 | -0.332 | 0.143 | -2.973 |
|
Marital status → resilience | -0.020 | -0.016 | 0.070 | -0.280 | -0.136 | -0.102 | 0.091 | -1.488 | 0.041 | 0.027 | 0.171 | 0.237 | |
Educational level → resilience | -0.023 | -0.017 | 0.072 | -0.321 | -0.268 | -0.167 | 0.096 | -2.784 |
-0.056 | -0.045 | 0.102 | -0.543 | |
Gender → burnout | -1.594 | -0.089 | 0.844 | -1.888 | -0.652 | -0.055 | 0.656 | -0.993 | 0.676 | 0.027 | 1.751 | 0.386 | |
Age → burnout | 1.074 | 0.087 | 0.652 | 1.647 | -0.806 | -0.077 | 0.703 | -1.147 | 0.108 | 0.009 | 1.197 | 0.090 | |
Marital status → burnout | 0.552 | 0.058 | 0.498 | 1.110 | 1.112 | 0.114 | 0.637 | 1.746 | 2.561 | 0.179 | 1.371 | 1.868 | |
Educational level → burnout | -0.781 | -0.072 | 0.516 | -1.516 | 0.715 | 0.061 | 0.680 | 1.051 | 0.672 | 0.057 | 0.815 | 0.825 | |
Stress → resilience → burnout | 0.138 | 0.041 | (0.071, 0.234) | 0.002 | 0.084 | 0.054 | (-0.025, 0.191) | 0.116 | 0.147 | 0.066 | (0.047, 0.307) | 0.008 | |
Goodness of fit | χ2 | 189.993 (p<0.001, df=60) | 151.399 (p<0.001, df=60) | 89.964 (p=0.007, df=60) | |||||||||
TLI | 0.894 | 0.892 | 0.927 | ||||||||||
CFI | 0.930 | 0.929 | 0.952 | ||||||||||
RMSEA | 0.073 | 0.075 | 0.062 |
p<0.05,
p<0.01,
p<0.001.
B: unstandardized estimate, β: standardized estimate, SE: standard error, CR: critical ratio, CI: confidence interval, df: degree of freedom, CFI: comparative fit Index, TLI: Tucker Lewis index, RMSEA: root mean square error of approximation
Goodness-of-fit index of invariance tests between the three groups
Goodness-of-fit Index |
||||||
---|---|---|---|---|---|---|
χ2 | df | p-value | CFI | TLI | RMSEA | |
Mental health worker–call center consultant | ||||||
Configural invariance | 240.572 | 64 | <0.001 | 0.932 | 0.952 | 0.054 |
Metrics invariance | 246.318 | 71 | <0.001 | 0.939 | 0.952 | 0.051 |
Scalar invariance | 253.973 | 78 | <0.001 | 0.945 | 0.952 | 0.049 |
Factor variance invariance | 234.763 | 81 | <0.001 | 0.932 | 0.939 | 0.053 |
Mental health worker–school counselor | ||||||
Configural invariance | 140.155 | 64 | <0.001 | 0.928 | 0.949 | 0.054 |
Metrics invariance | 147.643 | 71 | <0.001 | 0.935 | 0.948 | 0.052 |
Scalar invariance | 159.743 | 77 | <0.001 | 0.934 | 0.942 | 0.052 |
Factor variance invariance | 164.563 | 80 | <0.001 | 0.936 | 0.943 | 0.051 |
School counselor–call center consultant | ||||||
Configural invariance | 170.272 | 64 | <0.001 | 0.921 | 0.944 | 0.057 |
Metrics invariance | 186.212 | 71 | <0.001 | 0.926 | 0.942 | 0.055 |
Scalar invariance | 195.456 | 77 | <0.001 | 0.930 | 0.940 | 0.054 |
Factor variance invariance | 203.400 | 80 | <0.001 | 0.930 | 0.937 | 0.054 |
df: degree of freedom, CFI: comparative fit index, TLI: Tucker Lewis Index, RMSEA: root mean square error of approximation
Latent mean analysis between the three groups
Category | Latent mean difference/significant probability |
Cohen’s d | Latent mean difference/significant probability |
Cohen’s d | Latent mean difference/significant probability |
Cohen’s d | |||
---|---|---|---|---|---|---|---|---|---|
Mental health workers (N=270) | Call center workers (N=403) | Mental health workers (N=270) | School counselors (N=133) | School counselors (N=133) | Call center workers (N=403) | ||||
Stress | 0 | 0.150 |
0.68 | 0 | -0.121 |
0.53 | 0 | 0.274 |
1.10 |
Resilience | 0 | -0.406 |
1.77 | 0 | 0.334 |
1.24 | 0 | -0.707 |
2.83 |
Burnout | 0 | 2.405 |
3.67 | 0 | -1.518 |
1.95 | 0 | 3.349 |
4.19 |
p<0.05,
p<0.01,
p<0.001